Study On DSP Processing System Of Welding Seam Image

At present, welding manufactue is developing its full automatic, intelligent and robot technology along with the high development of computer technique. Automatic welding seam tracking is an important part of weld automation. To track welding seam automatically, a DSP processing system of welding seam image has been fulfilled and its template match algorithm has been build.The DSP processing system of welding seam image is composed of image capturing system, image processing system and image displaying system. Image capturing system finishes the task of capturing and converting image. It is composed of CCD and decoder. Image processing system finishes the task of image processing and identifying the position of welding seam. Its cord chip is DM642 DSP. Image displaying system finishes the task of displaying image processing result. It is composed of encoder and display.The template match algorithm is put forward to recognize the groove and seam center position, the algorithm is composed of recognizable texture pick-up, template building and template match. On recognizable texture pick-up, the seam image is transformed with wavelet and then processed with binary algorithm. Criterion for wavelet scale of recognizable texture pick-up is worked out. Through analyzing the groove and seam images, the texture pick-up method is concluded as that the groove images are only need to be transformed with horizontal wavelet and the average value of wavelet transform results' absolute values is taken as the threshold of binary processing ; the seam images are need to be transformed with horizontal and vertical wavelets and every two corresponding points of two wavelet transform results are squared respectively and added, then a new 2-D array is generated, the average value of the 2-D array is taken as binary process' threshold. On template building, the initial conditions of seam tracking are put forward, and then a groove or seam template is got according to the pipeline seam recognition initial conditions. On template match, first the examined image is transformed with 2D B-spline wavelet whose scale is identical with the scale used during template building, then the wavelet transform result is dealt with binary processing. Template is moved on the processing result of examined image to get according correlation coefficients, and then the seam center position is confirmed through template match criterion.At last, matching of groove and welding seam test are carried through. The result of test is analyzed. It is conclude that the DSP processing system of welding seam...